Skip to content

fix: allow mixed turbo/q8_0 KV cache types in CUDA flash-attention selection #12

fix: allow mixed turbo/q8_0 KV cache types in CUDA flash-attention selection

fix: allow mixed turbo/q8_0 KV cache types in CUDA flash-attention selection #12

Job Run time
20m 27s
18m 38s
32m 5s
1h 11m 10s